stream molecular testing during the diagnostic work-up of a suspected NSCLC if histology or cytology by itself cannot distinguish squamous cell carcinoma from adenocarcinoma. First, one marker for lung adenocarcinoma (transcription factor TTF-1) and one marker for squamous cell carcinoma (usually p63 or p40an isoform of p63) are employed [71]. If conclusive results are not obtained by these markers, second-­line markers for lung adenocarcinoma (aspartic proteinase Napsin-A) and for squamous cell carcinoma (cytokeratin 5/6) can be used. To reveal glandular differentiation, a mucicarmine histochemical stain can also be utilized. In cases with carcinoma metastatic to the lungs, clinical and radiologic correlation should always be used to adjust the immunohistochemical work-up, particularly if more lung-speci c markers are negative [2].

Diferent Techniques in Genotyping

As recommended by the current guidelines from IASLC and two international pathology societies: AMP and CAP, molecular analysis of all lung adenocarcinomas (including mixed tumors having adenocarcinoma component) in advanced stage should be performed for EGFR mutations by PCR-based techniques, and for ALK gene rearrangements by FISH assay or screening immunohistochemistry [20]. Another commonly used molecular analysis for lung cancer is testing for KRAS mutations which show resistance to tyrosine kinase inhibitors. Besides these main genomic targets, there is a growing list of less common driver mutations in lung adenocarcinoma such as ROS1 rearrangements, ERBB2 and BRAF mutations, MET ampli cation, etc. The increasing number of genomic targets for lung cancer and one-off testing approach in molecular analysis will result in the depletion of the cellular specimen although the cytopathologist can maximize cellularity of cell block and minimize loss from the specimen in the initial work-up. Consequently, multiplexed panels will be a must in the near future.

High success rates of molecular testing on small biopsy and cytology specimens have been

reported in many pertinent publications. Small biopsy specimens (including TBB or transthoracic CNB) and cell block specimens generally have comparable success in molecular analysis. Depending on the study parameters, success rate of molecular testing is 55–100% on small biopsy specimens, and 46–95% on transthoracic FNA or EBUS-TBNA cell block specimens [72–74]. Owing to the limited tumor cellularity in small biopsy or cytology specimens, failure rate of molecular testing on them is higher than that on larger surgical specimens [75]. As reported in a publication by the Lung Cancer Mutation Consortium, using an 8-gene panel testing, about 35% of cytology specimens and 26% of small biopsies can be inadequate for molecular analysis compared to only 5% of surgical resection specimens. However, if a specimen is adequate (suf - cient tumor cellularity) for molecular analysis, using cytology, small biopsy or surgical resection specimen has no effect on the performance of the subsequent molecular testing [76]. Thus, cytology or small biopsy specimens with adequate tumor cells are shown to be appropriate for molecular analysis as long as there is optimal pathologic work-up that minimizes tumor cell loss.

Methods to Overcome Challenges in Tissue Acquisition and Genotyping

Targeted treatment that is personalized based on molecular pro ling of advanced NSCLC can provide an objective response rate exceeding 50%, a progression-free survival of about 3 years and a median overall survival exceeding 6 years for patients with ALK rearrangements. Therefore, identifying molecular biomarkers is a requisite in advanced-stage NSCLC to tailorize the treatment aimed at optimal outcomes [77, 78]. However, there are several challenges in tissue acquisition, handling and processing for pathological diagnosis and genotyping:

1.Inadequate lung cancer tissue (10–20%) may be obtained by minimally invasive procedures and these procedures may need to be repeated,

2.Histological and biological heterogeneity of the tumor not captured in small biopsies or cytological specimens may negatively impact detection of speci c molecular targets. Different molecular pro les in primary tumor and metastatic lesion (intertumor heterogeneity), and alterations in the tumor biology due to treatment may require serial biopsies to track tumor evolution.

3.Resistance mechanisms may be heterogeneous in multiple metastases. Thus, re-biopsy of the progressive tumor site to clarify the resistance mechanism may not be representative. It may also not show the genetic heterogeneity of the whole tumor owing to intratumor heterogeneity.

4.Poor performance status of most lung cancer patients with advanced and/or recurring disease, and complication risks due to biopsy procedures reduce the feasibilty of tissue acquisition and genotyping.

These challenges can be potentially overcome with several approaches: rapid onsite evaluation, combination of minimally invasive procedures with sensitive genotyping assays, and/or liquid biopsy (analysis of tumor cell genomic contents in body fuids) [78–80].

Rapid on-Site Evaluation (ROSE)

ROSE, combined with cell block preparation, can optimize diagnostic performance of cytological specimens obtained by EBUS-TBNA and other procedures. It increases the diagnostic sensitivity by about 8% with no increase in procedure time. Furthermore, ROSE allows for repeating aspirations from sites giving diagnostic specimens. In an approach to triage small specimens appropriately, these additional specimens are separated for processing by priority to increase yields in histologic and molecular analysis. ROSE is recommended for procedures obtaining cytological specimens (e.g. EBUS-TBNA, FNA, touch imprints of CNB) but it may not be available in every institution owing to time, cost, and personnel limitations [4, 81, 82].

Diff-Quik smears for ROSE may be better than cell blocks in identifying genomic altera-

tions by NGS in lung cancer, and DNA can be extracted directly from ROSE cytology smears [78, 83]. Communication among interventionists, physician assistants, nurses, cytotechnologists and cytopathologists is critical. Furthermore, the whole team should use the same terminology for differentiating morphological adequacy from molecular adequacy [4, 82].

Sensitive Genotyping Assays

In scaling traditional single biomarker assays, tissue reduction in small biopsy or cytological samples is an increasingly encountered issue. Furthermore, the list of therapeutically relevant biomarkers have been expanding for NSCLC. Consequently, there has been a need for feasible and cost-effective assays characterizing a wider genomic pro le that may have prognostic and therapeutic implications.

Recently, there has been a strong tendency to use multiplexed genetic sequencing panels. NGS is at the forefront of this changing practice. Signi cantly higher sensitivity and speci-city of NGS compared with single-gene targeted assays has been shown previously. The optimal approach to molecular analysis in nonsquamous NSCLC is still under debate. Sequential, small-panel, or larger-panel NGS testing is suggested within potential strategies. Sequential testing is ­cost-­effective if employed only for EGFR, ALK and ROS1. However, upfront NGS becomes the optimal and costeffective strategy for an expanded panel beyond these three biomarkers [84, 85].

Besides EGFR, BRAF and MET mutations and ALK, ROS1, RET and NTRK translocations that have already been included in the NSCLC diagnostic standards in parallel with the entrance of their inhibitors into clinical treatment, there are emerging biomarkers such as KRAS G12C substitutions and HER2 activating alterations that are likely to be included in NSCLC guidelines after the approval of the corresponding drugs. In addition to the genetic analyses, analysis of PD-L1 protein expression is also performed in NSCLC to direct the use of immune checkpoint inhibitors. The integration of multiple molecular assays into a single diagnostic pipeline is the aim of ongoing studies. For com-